✨ feat: 添加在线API检查功能，更新配置项以支持最大音频长度和流式传输

Author: Cvandia

README.md

@@ -32,9 +32,6 @@ _⭐文本生成语音`tts`插件⭐_
 
 > 或者使用官方在线api -> [fish-audio](https://fish.audio/zh-CN/)即可享受快速云端的语音生成。
 
-> [!WARNING]
-> 由于目前国内无法正常访问fish-audio,需要配置[代理](#️-配置)或者使用离线fish-speech生成语音。
-
 ## 📜 免责声明
 
 > [!CAUTION]
@@ -91,6 +88,7 @@ git clone https://github.com/Cvandia/nonebot-plugin-fishspeech-tts
 | :--------------: | :-----: | :----: | :---------------: | :---------------------------------------------------------: |
 |  tts_is_online   |  bool   |   是   |       True        |                       是否使用云端api                       |
 | tts_chunk_length | literal |   否   |     "normal"      | 请求时音频分片长度，默认为normal，可选：short, normal, long |
+| tts_max_new_tokens   |   int   |   否   |       800       |                   最大音频长度，默认为800(大概6秒)                   |
 |  tts_audio_path  |   str   |   否   | "./data/参考音频" |            语音素材路径，默认为"./data/参考音频"            |
 |    tts_prefix    |   str   |   否   |       None        |                    触发前缀，默认为None                     |
 
@@ -107,6 +105,7 @@ ___
 
 |        配置项        | 类型  | 必填项 | 默认值  |                                                                                     说明                                                                                     |
 | :------------------: | :---: | :----: | :-----: | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
+| online_api_url      |  str  |   否   | "https://api.fish-audio.cn"| 官网api地址  |
 | online_authorization |  str  |   是   | "xxxxx" |                                                    官网api鉴权秘钥，详见[链接](https://fish.audio/zh-CN/go-api/api-keys/)                                                    |
 |  online_model_first  | bool  |   否   |  True   | 如果你想调用官方模型，通过自己的参考音频，定制角色音色，将此项设为`False`。当然，如果你没有准备参考音频，也会调用官网已经有的音色，具体详见[链接](https://fish.audio/zh-CN/) |
 |   online_api_proxy   |  str  |   否   |  None   |                                                                     代理地址，如：http://127.0.0.1:7890                                                                      |

nonebot_plugin_fishspeech_tts/__init__.py

@@ -7,6 +7,7 @@
 require("nonebot_plugin_alconna")
 
 from . import matcher as _match  # noqa
+from . import on_start_up  # noqa
 
 
 usage: str = """

nonebot_plugin_fishspeech_tts/config.py

@@ -8,16 +8,19 @@ class Config(BaseModel):
     # 基础配置
     tts_is_online: bool = True
     tts_chunk_length: Literal["normal", "short", "long"] = "normal"
+    tts_max_new_tokens: int = 800  # 大约6秒
     tts_audio_path: str = "./data/参考音频"
     tts_prefix: Optional[str] = None
+    tts_is_stream: bool = False  # 是否流式传输
 
     # 区分配置
+    online_api_url: str = "https://api.fish-audio.cn"
     online_authorization: Optional[str] = "xxxxx"
     online_model_first: bool = True
     # 设置代理地址
     online_api_proxy: Optional[str] = None
 
-    offline_api_url: str = "http://127.0.0.1:8080"
+    offline_api_url: str = "http://127.0.0.1:8000"
 
 
 config = get_plugin_config(Config)

nonebot_plugin_fishspeech_tts/fish_audio_api.py

@@ -25,7 +25,10 @@
 from .request_params import ChunkLength, ServeReferenceAudio, ServeTTSRequest
 
 is_reference_id_first = config.online_model_first
-online_api_proxy = config.online_api_proxy
+API_URL = config.online_api_url
+API_PROXY = config.online_api_proxy
+IS_STREAM = config.tts_is_stream
+MAX_NEW_TOKENS = config.tts_max_new_tokens
 
 
 class FishAudioAPI:
@@ -34,9 +37,9 @@ class FishAudioAPI:
     """
 
     def __init__(self):
-        self.url: str = "https://api.fish.audio/v1/tts"
+        self.api_url: str = API_URL
         self.path_audio: Path = Path(config.tts_audio_path)
-        self.proxy = online_api_proxy
+        self.proxy = API_PROXY
 
         # 如果在线授权码为空, 且使用在线api, 则抛出异常
         if not config.online_authorization and config.tts_is_online:
@@ -65,7 +68,7 @@ async def _get_reference_id_by_speaker(self, speaker: str) -> str:
         exception:
             APIException: 获取语音角色列表为空
         """
-        request_api = "https://api.fish.audio/model"
+        request_api = self.api_url + "/model"
         sort_options = ["score", "task_count", "created_at"]
         async with AsyncClient(proxy=self.proxy) as client:
             for sort_by in sort_options:
@@ -128,6 +131,8 @@ async def generate_servettsrequest(
             opus_bitrate=24,
             normalize=True,
             chunk_length=chunk_length.value,
+            max_new_tokens=MAX_NEW_TOKENS,
+            streaming=IS_STREAM,
             references=references,
         )
 
@@ -148,7 +153,7 @@ async def generate_tts(self, request: ServeTTSRequest) -> bytes:
                     AsyncClient(proxy=self.proxy) as client,
                     client.stream(
                         "POST",
-                        self.url,
+                        self.api_url + "/v1/tts",
                         headers=self.headers,
                         content=ormsgpack.packb(
                             request.dict(),
@@ -170,7 +175,7 @@ async def generate_tts(self, request: ServeTTSRequest) -> bytes:
             try:
                 async with AsyncClient(proxy=self.proxy) as client:
                     response = await client.post(
-                        self.url,
+                        self.api_url + "/v1/tts",
                         headers=self.headers,
                         json=request.dict(),
                         timeout=60,
@@ -191,7 +196,7 @@ async def get_balance(self) -> float:
         """
         获取账户余额
         """
-        balance_url = "https://api.fish.audio/wallet/self/api-credit"
+        balance_url = self.api_url + "/wallet/self/api-credit"
         async with AsyncClient(proxy=self.proxy) as client:
             response = await client.get(balance_url, headers=self.headers)
             try:
@@ -204,7 +209,7 @@ def get_speaker_list(self) -> list[str]:
         获取语音角色列表
         """
         return_list = ["请查看官网了解更多: https://fish.audio/zh-CN/"]
-        if is_reference_id_first:
+        if not is_reference_id_first:
             try:
                 return_list.extend(get_path_speaker_list(self.path_audio))
             except FileHandleException:

nonebot_plugin_fishspeech_tts/fish_speech_api.py

@@ -17,11 +17,16 @@
 )
 from .request_params import ChunkLength, ServeReferenceAudio, ServeTTSRequest
 
+API_URL = config.offline_api_url + "/v1/tts"
+PATH_AUDIO = Path(config.tts_audio_path)
+MAX_NEW_TOKENS = config.tts_max_new_tokens
+IS_STREAM = config.tts_is_stream
+
 
 class FishSpeechAPI:
     def __init__(self):
-        self.api_url: str = config.offline_api_url + "/v1/tts"
-        self.path_audio: Path = Path(config.tts_audio_path)
+        self.api_url: str = API_URL
+        self.path_audio: Path = PATH_AUDIO
         self.headers = {
             "content-type": "application/msgpack",
         }
@@ -66,11 +71,11 @@ async def generate_servettsrequest(
             normalize=True,
             opus_bitrate=64,
             latency="normal",
-            max_new_tokens=800,
+            max_new_tokens=MAX_NEW_TOKENS,
             top_p=0.7,
             repetition_penalty=1.2,
             temperature=0.7,
-            streaming=False,
+            streaming=IS_STREAM,
             mp3_bitrate=64,
         )
 
@@ -99,7 +104,9 @@ async def generate_tts(self, request: ServeTTSRequest) -> bytes:
             RequestError,
         ) as e:
             logger.error(f"获取TTS音频失败: {e}")
-            raise HTTPException("获取TTS音频超时, 你的接口配置错误或者文本过长") from e
+            raise HTTPException(
+                f"{e}\n获取TTS音频超时, 你的接口配置错误或者文本过长"
+            ) from e
         except Exception as e:
             raise APIException(f"{e}\n获取TTS音频失败, 检查API后端") from e
 

nonebot_plugin_fishspeech_tts/on_start_up.py

@@ -0,0 +1,21 @@
+from httpx import AsyncClient, HTTPStatusError
+from nonebot import get_driver
+from nonebot.log import logger
+
+from .config import config
+
+IS_ONLINE = config.tts_is_online
+API = config.online_api_url
+
+driver = get_driver()
+if IS_ONLINE:
+
+    @driver.on_startup
+    async def check_online_api():
+        """检查在线API是否可用"""
+        async with AsyncClient() as client:
+            try:
+                response = await client.get(API)
+                response.raise_for_status()
+            except HTTPStatusError as e:
+                logger.warning(f"在线API不可用: {e}\n请尝试更换API地址或配置代理")